The rush to higher performance computers is driven by the need for large-scale simulations in science and engineering. This need goes in hand with the demand for highly scalable numerical methods and algorithms that allow for the use of nowadays massively parallel systems or modern architectures (as many-core and hybrid platforms) and deliver the expected gain in performance. The scalability of these algorithms and methods is a key point to design or improve simulation software used in Computational Science and Engineering.
This topic aims at providing a forum where recent developments in parallel numerical algorithms and their implementation on current parallel architectures, including massively parallel systems as well as systems with accelerators, are discussed. This includes the algorithmic design, implementation details, as well as a performance analysis. The topic’s themes are given by the following (incomplete) list:
Focus
- Numerical linear algebra for dense and sparse matrices
- Solvers for partial differential equations, including domain decomposition and multigrid
- Integrators for ODEs and systems of ODEs
- Solvers for DAEs and systems of DAEs
- Optimization methods
- Methods for non-linear problems
- Methods for high-dimensional systems
- Analysis methods for large data sets
- Uncertainty quantification.
Committee
Chair: Matthias Bolten (Universität Kassel, Germany)
Local chair: Laurent Philippe (FEMTO-ST, France)
Peter Arbenz (ETH Zurich, Switzerland)
El Mostafa Daoudi (Université Mohammed Premier-Oujda, Morroco)
Maya Neytcheva (Uppsala University, Sweden)
Marian Vajtersic (Universität Salzburg, Austria)
